SAVE Subcommand (MULTIPLE CORRESPONDENCE command)

The SAVE subcommand is used to add the transformed variables (category indicators replaced with optimal quantifications) and the object scores to the working data file.

  • Excluded cases are represented by a dot (the sysmis symbol) on every saved variable.

TRDATA. Transformed variables. Missing values specified to be treated as passive are represented by a dot. Following TRDATA, a rootname, and the number of dimensions to be saved can be specified in parentheses (if the number of dimensions is not specified, all dimensions are saved).

OBJECT. Object scores.

MULTIPLE CORRESPONDENCE adds three numbers. The first number uniquely identifies the source variable names, the middle number corresponds to the dimension number, and the last number uniquely identifies the MULTIPLE CORRESPONDENCE procedures with the successfully executed SAVE subcommands. Only one rootname can be specified and it can contain up to three characters. If more than one rootname is specified, the first rootname is used; if a rootname contains more than three characters, the first three characters are used at most.

  • If a rootname is not specified for TRDATA, rootname TRA is used to automatically generate unique variable names. The formula is ROOTNAMEk_m_n, where k increments from 1 to identify the source variable names by using the source variables’ position numbers in the ANALYSIS subcommand, m increments from 1 to identify the dimension number, and n increments from 1 to identify the MULTIPLE CORRESPONDENCE procedures with the successfully executed SAVE subcommands for a given data file in a continuous session. For example, with two variables specified on ANALYSIS and 2 dimensions to save, the first set of default names, if they do not exist in the data file, would be TRA1_1_1, TRA1_2_1, TRA2_1_1, TRA2_2_1. The next set of default names, if they do not exist in the data file, would be TRA1_1_2, TRA1_2_2, TRA2_1_2, TRA2_2_2. However, if, for example, TRA1_1_2 already exists in the data file, then the default names should be attempted as TRA1_1_3, TRA1_2_3, TRA2_1_3, TRA2_2_3. That is, the last number increments to the next available integer.
  • Following OBJECT, a rootname and the number of dimensions can be specified in parentheses (if the number of dimensions is not specified, all dimensions are saved), to which MULTIPLE CORRESPONDENCE adds two numbers separated by the underscore symbol (_). The first number corresponds to the dimension number. The second number uniquely identifies the MULTIPLE CORRESPONDENCE procedures with the successfully executed SAVE subcommands. Only one rootname can be specified, and it can contain up to five characters. If more than one rootname is specified, the first rootname is used; if a rootname contains more than five characters, the first five characters are used at most.
  • If a rootname is not specified for OBJECT, the rootname OBSCO is used to automatically generate unique variable names. The formula is ROOTNAMEm_n, where m increments from 1 to identify the dimension number and n increments from 1 to identify the MULTIPLE CORRESPONDENCE procedures with the successfully executed SAVE subcommands for a given data file in a continuous session. For example, if 2 dimensions are specified following OBJECT, the first set of default names, if they do not exist in the data file, would be OBSCO1_1, OBSCO2_1. The next set of default names, if they do not exist in the data file, would be OBSCO1_2, OBSCO2_2. However, if, for example, OBSCO2_2 already exists in the data file, then the default names should be attempted as OBSCO1_3, OBSCO2_3. That is, the second number increments to the next available integer.
  • Variable labels are created automatically. They are shown in the Notes table and can also be displayed in the Data Editor window.
  • If the number of dimensions is not specified, the SAVE subcommand saves all dimensions.